Vous êtes sur la page 1sur 7

GEOINFORMATICS – Vol. I - Radar Remote Sensing - S.

Quegan

RADAR REMOTE SENSING


S. Quegan
Sheffield Centre for Earth Observation Science, University of Sheffield, U.K.

Keywords: Scatterometry, altimetry, synthetic aperture radar, SAR, microwaves,


scattering models, geocoding, radargrammetry, interferometry, differential
interferometry, topographic mapping, digital elevation model, agriculture, forestry,
hydrology, soil moisture, earthquakes, floods, oceanography, sea ice, land ice, snow

Contents

1. Introduction

TE SS
2. Basic Properties of Radar Systems
3. Characteristics of Radar Systems

S
4. What a Radar Measures

R
5. Radar Sensors and Their Applications

AP L
6. Synthetic Aperture Radar Applications
H O
7. Future Prospects
Glossary
C E
Bibliography
Biographical Sketch
E O–

Summary

Radar sensors transmit radiation at radio wavelengths (i.e. from around 1 cm to several
PL C

meters) and use the measured return to infer properties of the earth’s surface. The
surface properties affecting the return (of which the most important are the dielectric
M ES

constant and geometrical structure) are very different from those determining
observations at optical and infrared frequencies. Hence radar offers distinctive
perspectives on the earth. In addition, the transparency of the atmosphere at radar
SA N

wavelengths means that cloud does not prevent observation of the earth, so radar is well
suited to monitoring purposes. Three types of spaceborne radar instrument are
U

particularly important. Scatterometers make very accurate measurements of the


backscatter from the earth, their most important use being to derive wind speeds and
directions over the ocean. Altimeters measure the distance between the satellite platform
and the surface to centimetric accuracy, from which several important geophysical
quantities can be recovered, such as the topography of the ocean surface and its
variation, ocean currents, significant wave height, and the mass balance and dynamics
of the major ice sheets. The third and most versatile radar instrument is synthetic
aperture radar (SAR). A unique aspect of SAR is its ability to provide precise
measurements of surface displacement, using techniques known as interferometry and
differential interferometry. This capability, together with the fine spatial resolution of
the images and the provision of long temporal sequences undisturbed by cloud, has led
to SAR being applied in numerous fields, from topographic mapping to oceanography.

©Encyclopedia of Life Support Systems (EOLSS)


GEOINFORMATICS – Vol. I - Radar Remote Sensing - S. Quegan

1. Introduction

The word “radar” is derived from “radio detection and ranging,” which makes clear the
military context in which it was first developed. It involves the transmission of signals
at radio frequencies. These signals scatter from objects in the radar beam, and
information about these objects is gained by measuring the scattered radiation (bats
employ the same principle, but using sound rather than electromagnetic waves). For
military purposes (and also for many civilian tasks, such as air-traffic control or ship
monitoring), the objects of interest are usually human-made and relatively small, and
the problem is to detect them against a background of scatter from their surroundings
(commonly called “clutter”) and random system noise. However, from an
environmental point of view, the clutter is often the most interesting part of the return,
because the scatter from natural surfaces can yield information about a wide range of

TE SS
geophysical phenomena.

S
Systematic experiments, beginning in the 1960s with ground-based systems,

R
demonstrated the sensitivity of radar to phenomena such as sea state, soil moisture,

AP L
vegetation development, etc. This motivated the development of remote sensing radars
H O
designed to look down at the earth. The early systems were carried on aircraft, but the
event that really demonstrated the value of this technology was the Seasat satellite
C E
mission, in 1978. As its name implies, the purpose of Seasat was to gather information
about the oceans. To this end it carried three radar instruments: an altimeter, a
E O–

scatterometer, and a synthetic aperture radar (SAR) (these are still the most important
types of radar sensors for remote sensing; their principles are described in Section 5).
The Seasat mission lasted only 100 days, but it produced truly remarkable images,
including maps of the topography of the ocean surface, measurements of surface wind
PL C

speeds over the ocean, and unexpected signatures of internal waves within the ocean.
There were also interesting images of land surfaces, indicating the possibility of
M ES

measuring land processes.

The success of Seasat stimulated worldwide interest in satellite radar, and the last
SA N

decade of the twentieth century saw the launch of the European Space Agency (ESA)
ERS-1 and ERS-2 satellites (launched in 1991 and 1995 respectively), the Russian
U

Almaz, the Japanese JERS-1, and the Canadian Radarsat. The United States has not
launched any further satellite SARs for Earth observation, but carried out the
spectacularly successful Magellan SAR mission to Venus, which allowed the surface of
the planet to be mapped at unprecedented detail. American efforts in space-based radar
remote sensing have instead been through satellite altimeters (for example, TOPEX-
Poseidon, in collaboration with France), scatterometers (NSCAT, jointly with Japan,
and QuikSCAT), and systems carried on the space shuttle. The shuttle has helped to
carry the field forward by demonstrating the space deployment of techniques, like
polarimetric radar, and to carry out specific tasks, such as global topographic mapping
by using interferometry (the SRTM mission). In addition, the JPL/NASA AirSAR
airborne SAR system has played a major role in exploring the value of multiple
frequencies and polarisations for remote sensing.

Since the early 1990s we have had continuous observations by radar satellites, and there
have been major advances in understanding what they bring to our knowledge of the

©Encyclopedia of Life Support Systems (EOLSS)


GEOINFORMATICS – Vol. I - Radar Remote Sensing - S. Quegan

earth and its processes. Radar data have provided information on phenomena as diverse
as earthquakes, ocean currents, forest biomass, and the dynamics of glaciers. Many of
these applications exploit properties that are unique to radar systems, and that ensure the
place of radar in the range of sensors needed to understand properly the earth as a
system. To appreciate what makes radar special in the panoply of remote sensing
instruments, we first set out some of these properties. The basic terminology of radar
and types of radar are then explained, together with a non-exhaustive survey of the
many environmental applications where radar has contributed. The last decade of the
twentieth century taught us a lot about the role of radar in environmental remote
sensing, and what we have realised is that current sensors are often not ideal for many
purposes: understanding and the development of new ways to recover information from
radar have leapt ahead of the design cycle of satellite sensors. The last section of this
article therefore attempts to look to the future of radar, and the systems likely to be of

TE SS
most importance over the coming years.

S
2. Basic Properties of Radar Systems

R
AP L
As the name suggests, radar remote sensing systems operate in the radio (microwave)
H O
range of frequencies, from about 0.03 to 30 GHz. This is five to six orders of magnitude
less than those of the optical bands, and corresponds to wavelengths from about 1 cm to
C E
about 10 m. Many consequences flow from this, both for the measurements possible by
these systems and their response to the environment. Among these are:
E O–

• At these wavelengths, the atmosphere is essentially transparent, so the radar is


unaffected by cloud. Except at the shortest wavelengths, it is also unaffected by rain.
Hence radar can acquire images of Earth under all weather conditions. In a wider
PL C

context, the outstanding images of Venus provided by the Pioneer, Venera, and
particularly the Magellan radar satellites strikingly illustrate this cloud-penetrating
M ES

property. The planet Venus is completely cloud covered, and everything we know
about its surface comes from these radar images. A dramatic example is given in
Figure 1, derived from Magellan data, which shows a remarkable “half crater”
SA N

caused by the formation of a fault valley across the crater some time after an
asteroid impact.
U

Figure 1. Image of an unnamed crater located in the rift between Rhea and Theia
Montes in Beta Regio on the surface of Venus, derived from Magellan data
(Source: NASA/JPL/Caltech)

©Encyclopedia of Life Support Systems (EOLSS)


GEOINFORMATICS – Vol. I - Radar Remote Sensing - S. Quegan

• As wavelength increases, the ability of the radar to penetrate vegetation canopies


and into the soil also increases, giving the possibility of measuring soil properties
through overlying vegetation and observing sub-surface structures.
• Whereas optical sensors respond essentially to rotational and vibrational energy
bands of surface materials, radar is sensitive to geometrical structure and dielectric.
For many types of land cover, such as vegetation, the water content is a major
determinant of the dielectric, and changes in water content have large effects on the
properties of the returned signal. For example, the depth to which a radar signal can
penetrate into soil is very dependent on the soil moisture. In arid areas, long
wavelengths can penetrate many metres; this has been used, for example, to
delineate former drainage patterns under what are now sand deserts.
• The technology at these wavelengths allows the phase (see Section 3.
Characteristics of Radar Systems) of the returned signal to be measured. This has

TE SS
important consequences, because phase carries information about the nature of the
interaction with the earth’s surface. It is exploited to explore the polarisation

S
properties of the earth in radar polarimetry. In addition, differential phase (between

R
the returns to two antennae or to the same antenna at different times), measured at a

AP L
given sensor position, allows height measurements to be made. This is the basis of
H O
radar interferometry, from which digital elevation models and extremely precise
measurements of surface motion can be made.
C E
• Related to the ability to measure phase is the radar’s sensitivity to motion, through
the Doppler effect. This has both good and bad consequences: it allows inferences to
E O–

be made about the speeds of moving objects like ships or cars, but it also greatly
complicates the imaging of targets like the ocean surface, which are in constant
motion.
PL C

As well as their huge difference in frequency, radar systems differ from optical sensors
by being active—they transmit the energy used to illuminate the earth’s surface,
M ES

typically as short pulses of radiation, then measure the returned signal. This has two
major consequences:
SA N

• They are not dependent on solar illumination, so can operate just as well by day or
by night. Combined with their insensitivity to atmospheric conditions, this makes
U

radars well suited to multi-temporal observations.


• The emitted signal is precisely known, so calibration is relatively straightforward. In
addition, because there is normally no need to worry about the intervening
atmosphere, the measurements correspond directly to geophysical properties of the
surface. This is quite different from the situation at optical wavelengths, where
changes in both solar illumination and the atmosphere have big effects on what is
measured at the sensor, making conversion to geophysical units difficult.

3. Characteristics of Radar Systems

Before discussing what radar actually measures, it is important to have an idea of the
system properties affecting the radar signal. Designers of remote sensing radars must
make a variety of choices, depending on the intended application of the sensor (and, of
course, the funds available). The most important decisions concern the frequency,
polarisation, and incidence angle characteristics of the system, although also very

©Encyclopedia of Life Support Systems (EOLSS)


GEOINFORMATICS – Vol. I - Radar Remote Sensing - S. Quegan

important issues for applications are spatial resolution and time between successive data
acquisitions (the revisit time of the sensor). Two important descriptors of an
electromagnetic wave are its amplitude and phase. In its simplest terms, the electric
field measured at a fixed position for a pure sinusoidal wave has the form a cos (2πft +
ϕo), where f is frequency (in Hz) and t is time. The positive quantity a is known as the
amplitude of the wave and gives the maximum value of the electric field. The
expression (2πft + ϕo) is called the phase of the wave, with ϕ0 representing a phase
reference when t = 0.

3.1. Frequency

The choice of frequencies is determined not just by the desired properties of the sensor,
but also by the need to accommodate the needs of other users of the radio bands. This

TE SS
restricts the available frequency bands, which, as a hangover from World War II, are

S
often referred to by letters. The frequencies most commonly used by remote sensing
radars are given in Table 1, along with the corresponding wavelengths and examples of

R
systems using them. (Note that conversion between frequency and wavelength is easy:

AP L
frequency in GHz x wavelength in cm = 30.) Some of these systems used several
H O
frequencies: for example, the space shuttle SIR-C/X-SAR mission carried instruments
operating at X, C, and L bands, while the airborne NASA/JPL system operates at C, L,
C E
and P bands. At present, no P band instrument has been flown in space, although this
possibility is being actively pursued, because of its importance for measuring vegetation
E O–

biomass and thus contributing to carbon-cycle studies. Longer wavelengths would be of


even greater value for this purpose, as demonstrated by the CARABAS airborne radar
operating at VHF frequencies (3–15 m wavelength), but the technical problems of
deploying such an instrument in space are too great for this to be considered at present.
PL C

Frequency Wavelength
M ES

Band Sensor
(GHz) (cm)
K 15.0 2 ERS altimeter
X 10.0 3 SIR-C/X-SAR
SA N

C 5.0 6 ERS, Radarsat SARs


S 3.0 10 Almaz SAR
U

L 1.2 24 Seasat, JERS SAR


P 0.5 68 AirSAR

Table 1. Radar frequency bands commonly used for remote sensing, and examples of
systems employing them

3.2. Polarisation

Most remote sensing radars transmit linearly polarised signals. These are described as
vertically (V) polarised if the electric field of the wave lies in the plane defined by the
propagation direction and the direction of a line pointing outwards from the earth’s
centre (a radius vector), and as horizontally (H) polarised if the electric field is
perpendicular to this plane. The return signal typically contains both vertically and
horizontally polarised components, which can be measured separately by an

©Encyclopedia of Life Support Systems (EOLSS)


GEOINFORMATICS – Vol. I - Radar Remote Sensing - S. Quegan

appropriately designed system. The measured signals are then described as the VV, VH,
HV, or HH channels, where VH, for example, denotes V transmit, H receive. Some
systems use interleaved transmissions at H and V polarisations, allowing them to
measure all four channels. An important system characteristic is whether the phase
differences between the channels are preserved. If all four channels are measured with
their correct phase differences, the system is described as fully polarimetric (or partially
polarimetric if a single polarisation is transmitted but two are measured; for example,
only V may be transmitted but VV and VH may be measured). If the phase differences
are lost, the sensor is multi-polarised. Fully polarimetric sensors give a complete
description of the scattering properties of the surface at the probing frequency. An
important property of such data is that it allows reconstruction of the signal that would
be measured for any choice of transmitted and received polarisation configurations,
through a process known as polarisation synthesis.

TE SS
Examples of the polarisations and frequencies used by a variety of systems are given in

S
Table 2. Note that the Envisat ASAR (launched in early 2002) will not preserve phase

R
differences, so will not be polarimetric, whereas the Canadian Radarsat-2 (to be

AP L
launched in late 2005) and Japanese PALSAR (to be launched in late 2004) will both
H O
have polarimetric modes.
C E
Frequency
Sensor Polarisation Incidence angle
band
23o
E O–

ERS SAR C VV
ERS C VV 18o to 59o
scatterometer
Radarsat C HH Selectable: 20o to
50o
PL C

JERS L HH 35o
15o to 55o
M ES

SIR-C C, L Fully polarimetric


AirSAR C, L, P Fully polarimetric Approx. 20o to 60o
Envisat C HH & HV, VV & VH, or HH & VV Selectable: 14o to
ASAR 45o
SA N
U

Table 2. Frequencies, polarisations, and approximate incidence angles for selected


sensors

3.3. Incidence Angle

Incidence angle is the angle between the propagation vector of the radar wave and the
radius vector at the earth’s surface. For satellite SAR, the range of incidence angles
within a swath is normally only a few degrees, although Radarsat and Envisat have
selectable swaths, allowing a large range of angles to be covered. For airborne sensors,
the incidence angle normally varies considerably across the swath. This is important
because it can significantly affect the nature of the dominant scattering processes, and
hence the properties of the response. Table 2 gives the approximate incidence angles
associated with a variety of sensors.

©Encyclopedia of Life Support Systems (EOLSS)


GEOINFORMATICS – Vol. I - Radar Remote Sensing - S. Quegan

-
-
-

TO ACCESS ALL THE 26 PAGES OF THIS CHAPTER,


Visit: http://www.eolss.net/Eolss-sampleAllChapter.aspx

Bibliography

Elachi C. (1987). Spaceborne Radar Remote Sensing: Applications and Techniques, 254 pp. New York:
IEEE Press. [A readable account of the principles behind the main spaceborne radar instruments and their

TE SS
use in geophysical applications.]

S
Henderson F.M. and Lewis A.J., eds. (1998). Principles & Applications of Imaging Radar (Vol. 2 of The
Manual of Remote Sensing), 3rd ed., 866 pp. New York: John Wiley. [This source book covers most of

R
AP L
the main areas to which SAR has been applied. Several of the chapters are valuable both as overviews of
basic concepts and reviews of the literature, although other chapters are less useful, and several are not up
H O
to date.]
Kingsley S.P. and Quegan S. (1992). Understanding Radar Systems, 375 pp. London: McGraw-Hill. [A
C E
beginner’s guide to radar, including elementary descriptions of radar remote sensing systems.]
Oliver C.J and Quegan S. (1998). Understanding Synthetic Aperture Radar Images, 479 pp. Boston:
E O–

Artech House. [A comprehensive description of the properties of SAR images and methods of data
handling to recover information from them.]
Schreier G., ed. (1993). SAR Geocoding: Data and Systems, 435 pp. Karlsruhe, Germany: Wichmann
Verlag. [This collection of 18 papers covers many aspects of SAR images, in addition to their geometric
PL C

properties and the operations necessary to correct them geometrically.]


Ulaby F.T., Moore R.K., and Fung A.K. (1981, 1986). Microwave Remote Sensing: Active and Passive, 3
M ES

vols., 2160 pp. Norwood, Mass.: Artech House. [This is an outstanding text. Despite its age, it is the basic
reference on scatterometry and SAR, dealing with systems, physics, and applications, as well as giving a
good account of the historical development of the field.]
SA N

Ulaby F.T. and Elachi C., eds. (1990). Radar Polarimetry for Geoscience Applications, 364 pp. Norwood,
Mass.: Artech House. [A wide-ranging survey of polarimetric systems, polarimetric scattering theories,
U

and applications of polarimetry.]

Biographical Sketch

Shaun Quegan received the B.A. (Hons. First Class, 1970) in mathematics and M.Sc. (1972) in
theoretical statistics from the University of Warwick. He taught for a number of years, before undertaking
a Ph.D. concerned with atmospheric modelling, which was awarded by the University of Sheffield in
1982. Between 1982 and 1986 he worked as a research scientist at Marconi Research Centre, for the last
two years of which he led the Remote Sensing Applications Group. He established the SAR Research
Group at the University of Sheffield in 1986, whose success led to his professorship, awarded in 1993. In
the same year, he helped to inaugurate the Sheffield Centre for Earth Observation Science, of which he
remains the director. In 2001, he became director of a new National Environmental Research Council
Earth Observation Centre of Excellence in Terrestrial Carbon Dynamics, whose purpose is to give better
understanding and greater quantitative estimation of the role of terrestrial ecosytems in the earth’s carbon
cycle. He has great expertise in the physics, systems, and data analysis aspects of radar remote sensing,
but his current interests lie more in the exploitation of radar and other remote sensing technologies in
environmental science and land applications.

©Encyclopedia of Life Support Systems (EOLSS)